Abstract
In this paper, we identify the Bott connection on the natural foliation of the projective sphere bundle of a Finsler manifold to the Chern connection of this manifold. As a consequence, the symmetrization of the Bott connection turns out to be the Cartan connection of the Finsler manifold. Following Liu-Zhang, the Cartan connection can also be obtained through an adiabatic limit process. Furthermore, a Chern-Simons type form is defined and its conformal properties are discussed.
Introduction
In classical Riemannian geometry, the Levi-Civita connection provides a unique, natural method for parallel transport that is both torsion-free and metric-compatible. However, in Finsler geometry—where the metric depends on both position and direction—the landscape is more complex. Geometers must choose between several competing connections, such as the Chern, Cartan, or Berwald connections, each preserving different geometric properties.
This research bridges the gap between specialized Finsler geometry and the broader fields of foliation theory and index theory. By situating Finsler connections within the framework of the projective sphere bundle ($SM$), the authors demonstrate that these connections are not merely arbitrary choices but are deeply rooted in the fundamental geometric structures of foliated manifolds.
The Bott-Chern Correspondence
The central geometric object in this study is the projective sphere bundle $SM$ of a Finsler manifold $M$. This bundle possesses a natural foliation—a way of slicing the manifold into vertical fibers (the spheres). In the study of foliations, the Bott connection is a standard tool, typically defined to be flat along the leaves of the foliation.
One of the paper's primary contributions is the formal identification of the Bott connection on the horizontal distribution of $SM$ with the Chern connection. This result, formalized in Theorem 1, provides a clear, extrinsic interpretation of the Chern connection: it is the natural connection arising from the foliated structure of the sphere bundle. This identification allows techniques from foliation theory to be applied directly to Finslerian problems.
The Cartan Connection and Adiabatic Limits
While the Chern connection is recovered directly from the Bott connection, the Cartan connection requires a different approach. The authors show that the Cartan connection is the "symmetrization" of the Bott connection, achieved by incorporating the Cartan endomorphism tensor.
To provide a more physical or dynamical origin for the Cartan connection, the authors employ the adiabatic limit process. This technique involves rescaling the metric on the sphere bundle to effectively "shrink" the vertical fibers toward zero size. The paper proves (Proposition 1) that the Cartan connection is the precise limit of the standard Riemannian Levi-Civita connection under this rescaling. This establishes the Cartan connection as the most natural Riemannian-like object when viewing the sphere bundle from a distance, reinforcing its importance in the Finsler hierarchy.
Chern-Simons Type Forms and Conformal Invariants
Building on these connection identities, the authors introduce a new secondary characteristic class: a Chern-Simons type form defined as $\eta \wedge (d\eta)^{n-1}$, where $\eta$ is derived from the Cartan form.
In the context of 2D Finsler surfaces, the authors provide an explicit transgression formula, showing that this form relates to the underlying cohomology of the manifold. A significant portion of the analysis is dedicated to conformal invariance. The authors identify the specific conditions under which this Chern-Simons type form remains invariant when the Finsler metric undergoes a conformal scaling. This suggests that the form could serve as a powerful tool for distinguishing between different types of Finsler structures, particularly those that are non-Riemannian.
Conclusion and Future Directions
By identifying the Chern and Cartan connections as manifestations of the Bott connection and adiabatic limits, respectively, this work provides a unified perspective on Finsler geometry. It moves the field away from a collection of disparate definitions and toward a cohesive theory integrated with modern geometric analysis.
The authors suggest that future research should focus on the topological properties of the new Chern-Simons type form. Specifically, investigating manifolds where the exterior derivative of the Cartan form is nilpotent ($(d\eta)^k = 0$) may reveal new classifications for Finsler manifolds that occupy the space between Berwald-type and general Finsler structures.
Summary updated